An allosteric network governs Tom70 conformational dynamics to coordinate mitochondrial import

Plasmid/Construct Design

The gene encoding full-length human Tom70 was codon-optimized for E. coli expression and synthesized by BioBasic (Canada). Two constructs encoding the Tom70 cytosolic domain (residues 108–608; HsTom70c) were prepared by subcloning into modified pET-24a vectors: one with an N-terminal His₆–MBP tag followed by a Human Rhinovirus B14 3C (3CRV) protease cleavage site, and another with a C-terminal 3CRV cleavage site followed by a tandem StrepII–His₆ tag. The C-terminally tagged HsTom70c was used for crystallographic studies and Orf9b coexpression, while the N-terminally tagged HsTom70c was used for HDX-MS experiments of apoHsTom70c or Orf9b (44–70) peptide incubated HsTom70c to minimize potential dynamic artifacts from C-terminal fusion in the absence of full-length Orf9b.

For the generation of the HsTom70c-Orf9b complexes, a co-expression construct was prepared as previously described²⁵. The full-length Orf9b gene was codon-optimized for E. coli expression, synthesized by GENEWIZ, and cloned into multiple cloning site 1 (MCS1) of a pET-DUET1 vector. The Tom70 cytosolic domain was inserted into MCS2 of the same vector and fused to a C-terminal 3CRV cleavage site followed by a tandem StrepII–His₆ tag. All primers were synthesized by Integrated DNA Technologies and all cloning approaches utilized Gibson assembly methods³². Assembled plasmids were maintained in E. coli DH503b (Invitrogen) for propagation and all constructs were sequence-confirmed by GENEWIZ.

Recombinant Protein Expression and Purification

For crystallographic studies, the pET-24a vector encoding HsTom70c with a C-terminal 3C protease cleavage site and tandem StrepII-His₆ tags was transformed into E. coli LOBSTR³³ cells. Cultures were grown in terrific broth (TB) supplemented with 50 μg/mL kanamycin at 37°C until an OD₆₀₀ of ~0.6–0.8. Protein expression was induced with 0.5 mM IPTG, followed by incubation at 18°C for 16–18 hours. Cells were harvested by centrifugation, resuspended in ice-cold Nickel Buffer A (20 mM Tris-HCl (pH 8.0), 500 mM NaCl, 10% glycerol, 25 mM imidazole-HCl (pH 8.0), 5 mM β-mercaptoethanol (BME), and supplemented with 1 mM MgCl₂, 1 mM CaCl₂, 1 mM PMSF, 5 mM benzamidine, 10 μM leupeptin, 1 μM pepstatin A, 2 μg/mL aprotinin, and 0.1 mg/mL DNase I (Millipore Sigma). Cells were lysed by sonication on ice (Branson SFX550), and the lysate was clarified by centrifugation at 20,000 × g for 30 minutes at 4°C. All chromatography steps were carried out at 4°C, and SDS-PAGE was used to monitor elution fractions and assess protein purity.

The clarified lysate was loaded at 5 mL/min onto two tandem 5 mL HisTrap™ HP columns (Cytiva) equilibrated in Nickel Buffer A. Columns were washed with 50 column volumes (CV) of buffer, and bound protein was eluted with 10 CV of Nickel Buffer B (20 mM Tris-HCl (pH 8.0), 150 mM NaCl, 500 mM imidazole-HCl (pH 8.0), 5 mM BME). Eluted fractions were pooled and passed through 10 mL of Strep-Tactin Superflow resin (IBA) equilibrated in Strep Buffer A (20 mM Tris-HCl (pH 8.0), 150 mM NaCl, 5 mM BME). After washing with 25 CV of Strep Buffer A, the protein was eluted with 5 CV of Strep Buffer B (Strep Buffer A plus 2.5 mM desthiobiotin). Eluted protein was pooled and incubated overnight at 4°C with 0.1 mg/mL His₆-tagged 3CRV protease to remove the affinity tag. The reaction was adjusted to 500 mM NaCl and reapplied to the tandem HisTrap™ columns equilibrated in Strep Buffer A to remove the His-tagged protease and cleaved tags. The flowthrough, containing the target protein with a C-terminal 3CRV protease tag scar, was collected. This flowthrough was diluted with 20 mM Tris-HCl (pH 8.0) to reduce NaCl to ~75 mM, then applied to two tandem 5 mL HiTrap Q HP anion exchange columns (Cytiva) equilibrated in Q Buffer A (20 mM Tris-HCl (pH 8.0), 75 mM NaCl, 1 mM DTT). After washing with 5 CV of Q Buffer A, bound protein was eluted over a 20 CV linear gradient from 0% to 100% Q Buffer B (20 mM Tris-HCl (pH 8.0), 500 mM NaCl, 1 mM DTT).Selected fractions were pooled, concentrated, and loaded onto a HiLoad™ 16/600 Superdex™ 200 pg size-exclusion column (Cytiva) equilibrated in SEC buffer (20 mM Tris-HCl (pH 8.0), 150 mM NaCl, 0.5 mM tris(2-carboxyethyl)phosphine (TCEP). Homogeneous fractions were pooled and concentrated to ~10 mg/mL. Samples were either used immediately or flash-frozen in liquid nitrogen with 10% glycerol and stored at −80°C.

For HDX-MS experiments, a pET-24a vector encoding HsTom70c with a N-terminal His₆-MBP tag followed by a 3C protease cleavage site was used for the expression of apo-HsTom70c. The HsTom70c-Orf9b complex was expressed using a pET-DUET1 vector with Orf9b in MCS1 and HsTom70c in MCS2, bearing a C-terminal 3CRV cleavage site and tandem StrepII-His₆ tag; expression and subsequent purification followed previously described methods with minor modifications²⁵. Each construct was transformed into E. coli LOBSTR³³ cells. Cultures were grown in TB with 50 μg/mL kanamycin for the cultures containing pET-24a N-terminally tagged HsTom70c, and 100 μg/mL ampicillin was used for cultures harboring the pET-DUET1 vector co-expressing Orf9b and C-terminally tagged HsTom70c. Cultures were grown at 37°C until an OD₆₀₀ of ~0.6–0.8. Expression was induced with 0.5 mM IPTG, and cells were incubated at 18°C for 16–18 hours. Cells were harvested by centrifugation, resuspended in ice-cold Nickel Buffer A (as described above) with protease inhibitors and DNase I, and lysed by sonication. Lysates were clarified at 20,000 × g for 30 minutes at 4°C. All purification steps were carried out at 4°C, with SDS-PAGE used to monitor protein purity. The clarified supernatant was loaded at 5 mL/min onto a 5 mL HisTrap™ HP column (Cytiva) equilibrated in Nickel Buffer A. After washing with 50 CV, bound protein was eluted with 10 CV of Nickel Buffer B. Eluted fractions were pooled, treated with 0.1 mg/mL His₆-3C protease, placed into dialysis tubing (3.5 kDa MWCO) (Thermo), and dialyzed overnight against 1 L of dialysis buffer (20 mM Tris-HCl (pH 8.0), 150 mM NaCl, 5 mM BME. Following cleavage, the sample was adjusted to 500 mM NaCl and passed through two tandem 5 mL HisTrap™ columns equilibrated in dialysis buffer to remove the protease and tag remnants. The flowthrough, containing protein harboring either an N-terminal 3CRV protease tag scar for apo-HsTom70c or a C-terminal 3CRV protease tag scar for HsTom70c in complex with Orf9b, was diluted to ~75 mM NaCl and applied to a 5 mL HiTrap Q HP columns equilibrated in Q Buffer A (20 mM Tris-HCl (pH 8.0), 75 mM NaCl, 1 mM dithiothreitol (DTT). After washing with 5 CV, the protein was eluted using a 20 CV gradient from 0% to 100% Q Buffer B (20 mM Tris-HCl (pH 8.0), 500 mM NaCl, 1 mM DTT). Final fractions were pooled, concentrated, and subjected to size-exclusion chromatography on a HiLoad™ 16/600 Superdex™ 200 pg column equilibrated in HDX buffer (10 mM HEPES-NaOH (pH 8.0), 75 mM NaCl, 1 mM EDTA-NaOH (pH 8.0), 0.5 mM TCEP). Homogeneous fractions were concentrated to ~10 mg/mL and either used fresh or flash-frozen in liquid nitrogen with 10% glycerol and stored at –80°C.

Crystallization and Structure Determination of HsTom70c

For crystallization of HsTom70c, purified protein underwent lysine methylation by incubation with borane (50 mM final concentration) and formaldehyde (100 mM final concentration) for 1 hr at 4°C. The reaction was then quenched with 25 mM glycine (final concentration) for 30 min on ice. Methylated HsTom70c was then buffer-exchanged into crystallization buffer (25 mM Tris-HCl (pH 7.5), 200 mM NaCl, 5 mM MgCl₂, and 1 mM TCEP) and concentrated to 16 mg/mL. Methylated HsTom70c was mixed in a 1:1 ratio with well solution containing 0.1M MES (pH 6.5), 0.1 M sodium acetate, and 30% PEG2000 MME in a sitting drop format. Crystals appeared after 24 hr and were harvested directly from the crystallization drop and frozen in liquid nitrogen.

Data were collected on beamline 24ID-C at the Advanced Photon Source at Argonne National Laboratory. Data were processed using the RAPD (v2.0.1) data processing pipeline, which uses XDS³⁴ for data indexing and reduction, AIMLESS³⁵ for scaling, and TRUNCATE³⁶ for conversion to structure factors. We determined the structure using molecular replacement in PHASER (v2.5.0)³⁷ using an initial model generated by Alphafold2 (v2.3.2)³⁸. The model was manually rebuilt using COOT (v1.1.19)³⁹ and refined in phenix.refine⁴⁰ using positional, B-factor, and TLS refinement (Table 1).

Hydrogen-Deuterium Exchange Mass Spectrometry (HDX-MS)

All samples were prepared for HDX-MS at an initial concentration of 10 μM in HDX Buffer. For peptide mixing experiments, synthetic Orf9b (44–70) peptide (VWR) was resuspended in phosphate-buffered saline to a stock concentration of 2 mM. HsTom70c was maintained at 10 μM and Orf9b (44–70) peptide was mixed to a working concentration of 90 μM to ensure at least roughly ~90% of HsTom70c was in complex with peptide at equilibrium after dilution during exchange.

HDX-MS was performed at the Biomolecular and Proteomics Mass Spectrometry Facility (BPMSF) of the University California, San Diego, using a Waters Synapt G2Si system with HDX technology (Waters Corporation) according to methods previously described⁴¹. Briefly, deuterium exchange reactions were performed using a Leap HDX PAL autosampler (Leap Technologies, Carrboro, NC). D2O buffer was prepared by lyophilizing HDX buffer initially dissolved in ultrapure water and redissolving the powder in the same volume of 99.96% D₂O (Cambridge Isotope Laboratories, Inc., Andover, MA) immediately before use. Deuterium exchange was measured in triplicate at each time point (0 min, 0.5 min, 1 min, 2 min, 5 min). For each deuteration time point, 4 μL of protein was held at 25°C for 5 min before being mixed with 56 μL of D2O buffer. The deuterium exchange was quenched for 1 min at 1°C by combining 50 μL of the deuteration reaction with 50 μL of 3M guanidine hydrochloride, final pH 2.66. The quenched sample (90 μL) was then injected in a 100 μL sample loop, followed by digestion on an in-line pepsin column (Immobilized Pepsin, Pierce) at 15°C. The resulting peptides were captured on a BEH C18 Vanguard precolumn, separated by analytical chromatography (Acquity UPLC BEH C18, 1.7 μm 1.0 × 50 mm, Waters Corporation) using a 7–85% acetonitrile gradient in 0.1% formic acid over 7.5 min, and electrosprayed into the Waters Synapt G2Si quadrupole time-of-flight mass spectrometer. The mass spectrometer was set to collect data in the Mobility, ESI+ mode; mass acquisition range of 200–2000 (m/z); scan time 0.4 s. Continuous lock mass correction was accomplished with infusion of leu-enkephalin (m/z = 556.277) every 30 s (mass accuracy of 1 ppm for calibration standard).

For peptide identification, the mass spectrometer was set to collect data in mobility-enhanced data-independent acquisition (MSE), mobility ESI+ mode instead. Peptide masses were identified from triplicate analyses and data were analyzed using the ProteinLynx global server (PLGS) version 3.0.3 (Waters Corporation). Peptide masses were identified using a minimum number of 250 ion counts for low energy peptides and 50 ion counts for their fragment ions. The following cutoffs were used to filter peptide sequence matches: minimum products per amino acid of 0.14, minimum score of 6.8, maximum MH+ error of 6 ppm, and a retention time RSD of 5%. The peptides identified in PLGS were then analyzed using DynamX 3.0 data analysis software (Waters Corporation). The relative deuterium uptake for each peptide was calculated by comparing the centroids of the mass envelopes of the deuterated samples with the undeuterated controls following previously published methods⁴².

For all HDX-MS data, at least 3 technical replicates were collected. Data are represented as mean values +/− STD of 3 technical replicates, though the LEAP robots high reproducibility may result in obscured error bars. The deuterium uptake was corrected for back-exchange using a global back exchange correction factor determined from the average percent exchange measured in disordered termini of various proteins⁴³. Deuterium uptake plots were generated in DECA(v1.16)⁴⁴, and the data are fitted with an exponential curve for ease of viewing. Peptides as shown in coverage map figures are the real coverage map data.

System Preparation for Molecular Dynamics (MD) Simulations

The refined X-ray structures for the HsTom70c_open and HsTom70c_closed states were missing atoms in the α9-α10 loop (resides 272–296 for HsTom70c_open and 273–287 for HsTom70c_closed). An ensemble of 20 models was generated using Alphafold3⁴⁵ and these models were clustered using ChimeraX software (v1.9)⁴⁶. The lowest energy model from the highest populated cluster for the α 9- α 10 loop, comprising residues 266–296, were then fit and transplanted to each starting structure prior to molecular dynamics (MD) simulations. Each model was solvated with explicit TIP3 water molecules and the number of Na⁺ and Cl⁻ ions were adjusted to neutralize the system charge and final ionic concentration set to 75mM to match HDX-MS experimental conditions. The final systems comprised a total of 264,008 and 264,125 atoms for the HsTom70c_open and HsTom70c_closed, respectively, with orthorhombic periodic cells of 137 ų for both systems.

All-atom MD simulations were performed using the CUDA memory-optimized version of NAMD 2.14⁴⁷ and CHARMM36m force fields⁴⁸. Each simulation began with an energy minimization step to remove any steric clashes and ensure that the system was in a low-energy state. A minimization of 2000 steps with the following key settings: a cutoff of 12.0 Å for nonbonded interactions, a pair-list distance of 14.0 Å was performed and the Particle Mesh Ewald (PME) method for long-range electrostatics with a grid spacing of 1 Å was used⁴⁹. A switch distance of 10.0 Å was enabled and all atoms were wrapped to the primary simulation cell, including water molecules and ions, to maintain periodic boundary conditions. Following minimization, the system underwent a temperature annealing process by gradually increasing the temperature from 60 K to 300 K using Langevin dynamics with a damping coefficient of 1 ps⁻¹ throughout the annealing phase. Each system was equilibrated at 300 K for a duration sufficient to reach a stable state. The equilibration phase used the same cutoff, pair-list distance, and PME settings as the previous steps to maintain consistency in the treatment of nonbonded interactions. Langevin dynamics continued to control the temperature, with a target pressure of 1.01325 bar maintained via a Langevin piston. The simulation time for this phase was 500,000 steps for both structures to ensure comprehensive equilibration. Harmonic constraints were applied during minimization, annealing, and equilibration to maintain the structural integrity of the starting complexes. Following minimization, heating, and equilibration, the systems were submitted for production MD simulations under NPT conditions. The MD production was carried out for 50,000,000 steps, equivalent to 100 ns, with coordinates saved every 1000 steps (2 ps). Randomized triplicates were conducted for each structure, yielding a combined total of 0.3 μs (150,000 frames) per structure. The first 10 nanoseconds of each trajectory were excluded to ensure equilibrated data for all subsequent analyses (Supplementary Figure 4).

Essential Dynamic Analysis (EDA) of MD Trajectories

To identify the predominant modes of motion observed during molecular dynamics simulations, Essential dynamic analysis (EDA)⁵⁰ was performed using the three 100-nanosecond concatenated trajectories from each system. The covariance matrix calculation used only protein C_α atomic coordinates and was analyzed using the ProDy software package (v2.4.1)⁵¹. The covariance matrix of Cartesian coordinates was created from position fluctuations after subtracting the mean structure. Eigenvalues and eigenvectors were obtained by diagonalizing the covariance matrix. Top eigenvectors were used to create dynamic animations and plots, showing correlated motions and major structural deformations.

Dynamical Network Analysis (DNA)

DNA was performed following the methodology described by Luthey-Schulten and colleagues³¹. Analyses were carried out using a Python-based framework (Python v3.9), employing the DNA package (v2.0) and MDAnalysis (v2.9.0)^52,53 for trajectory processing. Atom ordering consistency and periodic boundary condition removal were achieved using the PrepareSystem function from the DNA package. For each system, the three 100-nanosecond trajectories were concatenated and divided into four consecutive 75-nanosecond windows. The correlation network for each window was derived from a subsample of ten uniformly spaced frames. Nodes were defined as C_α atoms for amino acid residues and single pseudoatoms assigned to water molecules and ions where applicable. Residue pairs were connected by edges if the heavy atom distance between nodes remained within 4.5 Å for at least 75% of the simulation frames. Edge weights were calculated based on generalized correlation coefficients derived from mutual information estimates using Shannon entropy, with numerical performance optimized through NumPy (v2.3.1) and Numba (v0.61.2)⁵⁴.

Network visualizations were generated in Visual Molecular Dynamics (VMD, v1.9.4–57a)⁵⁵ using the custom NetworkView 2.0 graphical interface to highlight key residues and communities for clarity. The resulting visualizations were exported to UCSF ChimeraX (v1.9)⁴⁶ for final stylization and figure preparation.

HDX-MS-Guided Ensemble Refinement

To refine atomistic MD trajectories against our HDX-MS data, we used the HDXer open-source package (v1.2)³⁰. This approach employs a maximum-entropy reweighting formalism, which minimally adjusts the weights of each MD frame to achieve optimal agreement between back-calculated and experimental deuterium uptake values. This process selects for a subensemble of conformations that is most consistent with the experimental data while introducing the least possible perturbation to the original conformational distribution. The prior conformational ensemble was generated from all six independent, all-atom MD simulations (three closed-state, three open-state; 100 ns each). Solvent and ions were removed, and the coordinates were superposed on the reference crystal structure (C_α atoms using MDTraj (v1.9.9)⁵⁶. The dataset was downsampled to every fourth frame of the combined trajectory, yielding an input ensemble of 75,000 structures. The N-terminal methionine and the first amide of each peptide segment were excluded from the analysis to account for rapid back-exchange artifacts. The target dataset was derived from HDX-MS measurements performed at four exchange times (0.25, 0.5 1.0, and 2.0 mins). Back-exchange corrected deuterium uptake values were then converted to experimental protection factors under the assumption of EX2 exchange kinetics. For each frame in the input ensemble, theoretical protection factors were estimated using the empirical solvent-accessibility model implemented in HDXer. The reweighting was performed by using a hyper-regularization strength of γ equal to 2.0, a value determined from the “elbow” of an L-curve plot (Supplementary Figure 5). This value optimally balances agreement with experimental data and preservation of ensemble diversity. The initial predictions from the unbiased simulation showed qualitative agreement but also notable deviations from the experimental measurements across the four labeling timepoints (Supplementary Figure 6). As shown by the reweighted trace, the reweighting procedure yielded a better fit to the experimental deuteration fractions for most residues and timepoints, confirming that the refined ensemble is a more accurate representation of the protein’s solution-state dynamics.

To characterize the structural ensemble that best fits the experimental HDX data, conformations corresponding to the top 1% of the final HDXer weights were selected for further analysis. This high-weight subset of frames was partitioned into two distinct conformational states using K-means clustering based on the C_α atomic coordinates.

To visualize the relationship between these states, PCA was subsequently performed on the Cα coordinates of the same set of frames. The structures were projected onto the first two principal components (PC1 and PC2), and the conformational space occupied by each cluster was delineated with a convex hull. For each cluster, a representative centroid structure was saved, defined as the frame with the minimum RMSD to the geometric center of its respective cluster. Additionally, the single conformation corresponding to the highest individual weight in the reweighted ensemble was identified and saved for further inspection.